Econ 184b, Econometrics, Fall 2021
Module 0: Kickstarter
Xinde “James” Ji
Department of Economics, Brandeis Univeristy
xji@brandeis.edu
Office: Sachar 001A
Office hours: Wednesday 2:00-5:00 pm
Here’s where my office is:
- Xinde “James” Ji
- James / Professor Ji, he/him/his
- I speak Chinese (Mandarin, Shanghainese) and English
- I’m currently with my wife Sophia, and our 2-month-old daughter Reilly
- Game of Thrones / A Song of Ice and Fire - George RR Martin
- River Town - Peter Hessler
- Expecting Better - Emily Oster
Here are two truths and a lie about me:
- I can recite \(\pi\) to ~60 digits.
- I completed 1-12 grades in 10 years
- I can stand upside down and split out jellybeans
Why econometrics?
- ECON 80a teaches us about individual and firm behaviors
- Adam Smith’s invisible hands
- ECON 82b teaches us about how the macro-economy works
- Keynes, Friedman, rational expectation
- Why is econometrics a required course for the ECON major?
- Is it there to trick you folks?
Economics as a science
- “If you ask 10 economists about their forecasts for next year, there will be 12 different answers.”
- And economics was partly like astrology: make up some theory about how the world works, and be done with it
- Adam Smith: The invisible hand always adjusts
- Karl Marx: Price has to fluctuate around the cost of material/labor inputs
- Malthus: We will run out of food because land grows arithmatically, and population grows exponentially
Economics as a science
- What makes a discipline science?
Modern Physics
And now we know that
\[h = \frac{1}{2}gt^2\]
with the assumption that there is no air resistance.
(Of course, air resistance is also something that Newtonian mechanics can model.)
Galileo’s experiment
- Hypothesis: ligher ball falls as fast as heavier balls
- Experiment: drop two balls from the Pisa tower
- Data: time used with different initial heights
- Model: the free-fall equation
- Insights: the parameter g (the gravity constant)
Now, what is econometrics, again?
- Econometrics is the scientific process of economics
- Uses data to test hypothesis, generate insights
Question: How much is the return to education? Does that cover the actual/opportunity cost of going to college?
Theory:
- Education provides useful knowledge and skills, thus increasing later-life wages (the human capital hypothesis)
- More schooling is a signal for innate ability, as workers with higher ability find schoolwork easier (the signaling hypothesis)
- Individuals pays for college tuition, and forego ~4 years of wages by attending college
Now we do what?
Data on schooling and wage earnings
- Wage positively correlated with schooling
- One additional year of schooling increases hourly wage by ~10%
- Figured out using “regression”
- What could be wrong with this relationship?
What could be wrong?
- What about innate ability?
- Education could just be a signal rather than bringing actual gains in human capital
- Correlation != Causation
- And that’s where econometrics differ (a bit) from statistics and data analytics
- Ed Leamer(1983)’s critique
- Traditional econometric models are not trustworthy
- "Hardly anyone takes data analysis seriously
The credibility revolution
- That is where conometrics diverges traditional statistics/machine learning
- Instrumental variables
- Panel models
- Quasi-experimental approach
- Randomized control trials (RCT)
And it is not just economics
Yiqing Xu, Professor of Political Science at Stanford University:
“The field of political science has been modernized in the recent years by borrowing from economics - not only information asymetry, transaction cost … But also the quantitative methods in econometrics.”
And in other fields
- Public health
- History
- Tech firms
A roadmap of what we will be doing
- Theory and intuition of econometric tools
- Bivariate/multivariate regression
- Hypothesis testing
- IV, Panel data, quasi-experiments
- Implementation of these tools
- Use these tools to evaluate real-world questions
